Or use OCAML (if you want objects) or some other ML variant if you're not an OOP fanatic (Yes I know OOP is cool, but when you have polymorphism and a sane module system there aren't too many remaining benefits)
Every user you know? In our lab we have an OS-X laptop donated by a student who left recently. We all got a chance to try it, but nobody has switched. The Ti-book now sits idle waiting for the next person to try it.
One of the givaways IMNSHO is that people always compare OS-X to Gnome and KDE, as though those are the only things to run on Linux. Serious Linux/Unix users often use less pretty but more usable, far more configurable, and faster systems to get real work done.
I don't really care if the Linux world loses the people who installed one version of RedHat and never booted into it anyway. If you don't want to actually tweak your system or expect great throughput or practical networked or headless operations, perhaps OS-X is better for you anyway...
Re:Warping the definition of robot
on
Immobile Robots
·
· Score: 1
Well in that case "agent" would be a much better word. An agent is a situated entity that observes an environment, makes decisions, and takes actions in that environment. Note that this nicely includes active software systems, "immobots", and mobile robots, while ruling out things that just observe and require a human to take action, like a fault detector. Unfortunately the meaning of agent has been so dilluted by now that they decided to come up with something new. Immobot sounds pretty strange to me though...
Okay so I'm not completely unbiased, being a mobile robot researcher and all, but hey...
Perhaps they could call it "Phoenix, AZ". Not a catchy name perhaps, but a way to escape a silly lawsuit or settlement. Phoenix Technologies isn't even from Arizona anyway. I guess they might have to clear it with the city, but perhaps it could become Phoenix's official web browser:)
The next time I need to start a company I'll make sure to trademark "New York", "Seattle", or "Los Angeles" as part of the name. While we're at it, I guess its time for Mosaic Technologies, Operan Technologies, and Conquerer to settle accounts. I mean, look at all the confusion those naming conflicts have created!
Ok my previous post was probably a bit harsh, as most robot people will only have a background with wheeled robots. Well to make a long story short, walking robots still suck and wheels are still far more practical for almost all tasks. Locomotion demos often try to hide the thorny problems and make it look like walking will soon be practical, but its still quite far off IMO, though progress is being made. Legged robots mostly exist thanks to their coolness factor, which is sad given that a wheeled robot can be cheaper while traveling at 5-10 times the speed...
There are no truly dynamic non-planar walkers right now, and just adding sensors doesn't fix the problem; it's still extremely difficult. The nice thing about a quadruped is that the walk doesn't necesarily need to be feedback based (Oscillator designs are nice but they have yet to yield a practical 3d walker). The main problem is still actuator strength (as postulated by the MIT leg lab), not really balance or sensors.
That said, I'm not so impressed by 15m/min. We managed to get Sony Aibos to walk at 13m/min at about one fourth the scale of this thing. Our code is available for people to develop using the OpenR SDK (see https://www.openr.org/page1_2003/), so you could make your own security bot for ~$1500 instead of whatever this thing costs. It doesn't have a nose, but I guess you could always just keep the batteries fresh in your smoke detectors...
Gee, if only there was a shell that combined the strengths of bash, tcsh, and ksh, and brought them together as a powerful, consistent feature set. Oh wait, that's what the parent post was talking about: Zsh Rules!:-)
Seriously though, everyone should avoid being lazy and give it a try. Bash users would not even notice the difference, except that tab completion would work better. ex-tcsh users have to get used to some of the sh-isms (such as "export"), but again you don't lose any functionality while gaining a lot. And its not as bloated as one might first think; Its memory footprint is on par with bash.
Umm... no, that analogy doesn't work. If your blood is boiling, you die. The boiling temperature of water is less than room temperature in even a partial vacuum. Or through willpower do you expect to hold your breath against the 2 tons of force pushing out on your lungs? Meanwhile, your eyes would be boiling/exploding. So why does diving work? Simple, the air inside you compresses until there is no longer a pressure differential. The same happens in a vacuum, but by the time there is no longer a pressure differential in this case you would be dead.
But when you look at the numbers another way, calculating an individual person's likelihood of being involved in a fatal automobile accident in his or her lifetime, the percentages come out very close to zero.
Well, it is zero. You weren't involved in a fatal accident until you die, in which case it wasn't during your lifetime.:-)
This system could be taken out with a fairly ordinary power drill and a long bit, drilling one hole in the tube. Even just the partial loss of vacuum would be a huge problem for such high speed carriages.
Taking out a bridge with ordinary tools would be quite a bit harder (esp since its users could most likely see you doing it).
Using only metric simply garuntees that you'll be off by convenient powers of 10 when you violate your interfaces. If the interface checked the units, then in either case you would have easily found a problem.
In the robotics software my group writes, we've really had to drill it in to people to use the same metric unit (mm, sec, everywhere). In the past some of the modules differed, and had all sorts of ulgy *10/10, etc which could easily be forgotten and lead to bugs. We still have a platform that uses int usec for times, and occasionally still get bugs due to this (telling the difference between 10000000 and 1000000 while writing or scanning over code is easy to mess up).
In short, merely using metric doesn't garuntee the nonexistence of unit bugs, and the decrease in bugs due to things like inch/cm is offset by the fact that people just *know* "we all use metric so it will work" and thus are lazy in checking the order of the units "{mm,cm,m,km} is obviously the most logical since it makes the most sense for my module!". Common interfaces still need to be established and conformed to, that is the root problem.
The Sony robot is the SDR-4X. It's a lot more capable (it can actually walk right now), and a lot more expensive (though neither robot is being sold yet).
Too bad the user has to be running VirusScan or some other badly coded service for it to work. Funny how clueless nix users can be about their "enemy".
You seem to have missed the *whole point* of the original article. Go read it, and you'll find out any program with a text box will do. Poor coding has nothing to do with it, the bugs are in the Win32 API itself.
The strange thing is, he's not actually arguing that at all, and after criticizing things like the DMCA, goes on to say things such as:
Fortunately the technology itself - in the form of trusted computer architectures, secure networks and digital rights management - can be used to rescue the Net from US control.
I'm glad DRM can "rescue" us from the DMCA and evil corporate influence... They must take back the free network they had... and regulate it. yeah...
Well, it seems to be at least somewhat related to the Software Engineering Institute / CERT, etc. which is fairly separate from the rest of CMU, and a lot more large-software-corporation oriented than many of the research groups are. Of course, one of the nice things about a university is that it can be large enough to support both open communities and proprietary ones. Instead of trying to *fix* this one however, why not just support other open security projects, such as Ballista (which incidentally is also at CMU).
We have a new omni design as well, and it looks like the finalized robot (with all the junk on it) is going to be pretty fast. We'll have to watch out for the FU-Fighters, I hear they have an 8m/s (unconfirnmes) kicker they used at the German open. Of course kicking speed was not their problem in the past, but still... This year is going to be fun.
P.S. It was no fair not showing us your new robots when we visited:P (we didn't have ours yet). Are the new robots strong enough you can you stand on their covers like the 2000 robots?
That would be the FU-Fighters 1999 team. It was a spinning 5mm thick aluminum plate, which is pretty large considering the robot itself was about 10cm x 12cm. They could kick the ball at up to 6m/s. They turned it down later in the competetion, but it was not banned (more to keep from getting penalties than other team's anger).
The design lives on still in a toned down version on the current FU-Fighters team and others have copied it too. It's affectionately known as a "spinning blade of death" kicker.
(Note: I'm sitting in a RoboCup lab right now, so IMNSHO:)
I don't think the favorite is going to be Iran this year, but more likely the Phillips professional team, which won the German Open this year. That said, I wish people would realize there are 4 leagues, not just the middle size league, with different robots and different favorites in each. In the Sony Legged league, UNSW has dominated, though we came in second:) There are a lot of strong teams in the league though so we'll have to see...
In the small size, I'd say the favorites are last year's winner LuckyStar II from Singapore, and Big Red from Cornell University. FU-Fighters is also a pretty strong team. Our team (CMU) hopes to do a lot better this year in the small size league. We won in '97 and '98, but haven't done too well since then.
I don't know to much about the simulation league so I won't bother to comment. Finally, a personal plug: See a video from the vision system of a Sony legged robot here. It'll give you more respect for how hard a problem this is:) It sure did that for me, even though I've been programming them for several years.
None of it has anything to do with intelligence. Again, to be intelligent, a program must not only be able to learn but it must learn anything and everything, not just a limited domain environment. But that is not all. It must be motivated and adapt to reward and punishment.
Any of course, a human *can* learn "anything and everything"? Don't make me laugh. Humans are actually quite limited, just in different ways than computers. There are tons of things that already come up in AI that a human can't learn. This doesn't seem to stop people from saying a computer isn't intelligent until it can learn everything, or in other words be far superior to a human.
Netscape looked like it was going to die, so they started Mozilla, which I'm using happily right now. there aren't too many examples, but there are some...
Why did I buy an Nvidia video card? Well, because I actually did research rather making vague comparisons. As always, a well set up system with the best drivers will beat one where the person didn't even get agp enabled. If not, then explain why I should trust your opinions over a well known review site...
Yes, their lack of source is annoying. But the fact that they care about Linux as a platform (more than any other card manufacturer) means a lot to me. I know there are people at Nvidia paid to write those drivers, and as long as they have the fastest hardware too, they're likely to always be ahead of any other hardware/driver combination in Linux.
Ethernet is not expensive. They could probably add it for $10 or less. As others have pointed out, they'd save the money that it costs them for dialup for users that already have broadband. They could even do cooler "enhanced functionality" things with the software since it'll have a continuous connection instead of an intermittent one.
That said, I guess I can understand why it doesn't have an ethernet port yet. I expect to see one reasonably soon though.
One of the things that makes the X-box cool is that it does have an ethernet port built in; It allowed my brother and me to play Halo with others over his cable modem for free - That's a definite edge over the PS/2, and that feature didn't cost them anything besides ethernet support (they didn't even write the tunnel software, it was 3rd party).
I agree completely. Working with robots and attending public competitions every year (I'm on a RoboCup team), the topic of beam robots and Tilden's work invariably comes up at some point. People need to know that while interesting, this work doesn't scale. Right now, nothing scales very well; But competing to see how efficiently you can mimic something dumber than a sea-monkey doesn't make much of a difference. Tilden probably understands this, but of course he never says this in interviews, which is a shame.
So, I usually tell people this: Rodney Brookes' robots did pretty much the same thing in the 80s. These analog robots are cheaper and simpler, but no smarter (and usually dumber). I will believe they are the future of robots when they can beat the best computer controlled robot team at any nontrivial robotics competition (AAAI, RoboCup, Mirosot, etc.)
Slashdot Mirror
Or use OCAML (if you want objects) or some other ML variant if you're not an OOP fanatic (Yes I know OOP is cool, but when you have polymorphism and a sane module system there aren't too many remaining benefits)
Every user you know? In our lab we have an OS-X laptop donated by a student who left recently. We all got a chance to try it, but nobody has switched. The Ti-book now sits idle waiting for the next person to try it.
One of the givaways IMNSHO is that people always compare OS-X to Gnome and KDE, as though those are the only things to run on Linux. Serious Linux/Unix users often use less pretty but more usable, far more configurable, and faster systems to get real work done.
I don't really care if the Linux world loses the people who installed one version of RedHat and never booted into it anyway. If you don't want to actually tweak your system or expect great throughput or practical networked or headless operations, perhaps OS-X is better for you anyway...
Well in that case "agent" would be a much better word. An agent is a situated entity that observes an environment, makes decisions, and takes actions in that environment. Note that this nicely includes active software systems, "immobots", and mobile robots, while ruling out things that just observe and require a human to take action, like a fault detector. Unfortunately the meaning of agent has been so dilluted by now that they decided to come up with something new. Immobot sounds pretty strange to me though...
Okay so I'm not completely unbiased, being a mobile robot researcher and all, but hey...
Perhaps they could call it "Phoenix, AZ". Not a catchy name perhaps, but a way to escape a silly lawsuit or settlement. Phoenix Technologies isn't even from Arizona anyway. I guess they might have to clear it with the city, but perhaps it could become Phoenix's official web browser :)
The next time I need to start a company I'll make sure to trademark "New York", "Seattle", or "Los Angeles" as part of the name. While we're at it, I guess its time for Mosaic Technologies, Operan Technologies, and Conquerer to settle accounts. I mean, look at all the confusion those naming conflicts have created!
Ok my previous post was probably a bit harsh, as most robot people will only have a background with wheeled robots. Well to make a long story short, walking robots still suck and wheels are still far more practical for almost all tasks. Locomotion demos often try to hide the thorny problems and make it look like walking will soon be practical, but its still quite far off IMO, though progress is being made. Legged robots mostly exist thanks to their coolness factor, which is sad given that a wheeled robot can be cheaper while traveling at 5-10 times the speed...
Don't work on robot locomotion, do you?
There are no truly dynamic non-planar walkers right now, and just adding sensors doesn't fix the problem; it's still extremely difficult. The nice thing about a quadruped is that the walk doesn't necesarily need to be feedback based (Oscillator designs are nice but they have yet to yield a practical 3d walker). The main problem is still actuator strength (as postulated by the MIT leg lab), not really balance or sensors.
That said, I'm not so impressed by 15m/min. We managed to get Sony Aibos to walk at 13m/min at about one fourth the scale of this thing. Our code is available for people to develop using the OpenR SDK (see https://www.openr.org/page1_2003/), so you could make your own security bot for ~$1500 instead of whatever this thing costs. It doesn't have a nose, but I guess you could always just keep the batteries fresh in your smoke detectors...
Gee, if only there was a shell that combined the strengths of bash, tcsh, and ksh, and brought them together as a powerful, consistent feature set. Oh wait, that's what the parent post was talking about: Zsh Rules! :-)
Seriously though, everyone should avoid being lazy and give it a try. Bash users would not even notice the difference, except that tab completion would work better. ex-tcsh users have to get used to some of the sh-isms (such as "export"), but again you don't lose any functionality while gaining a lot. And its not as bloated as one might first think; Its memory footprint is on par with bash.
Umm... no, that analogy doesn't work. If your blood is boiling, you die. The boiling temperature of water is less than room temperature in even a partial vacuum. Or through willpower do you expect to hold your breath against the 2 tons of force pushing out on your lungs? Meanwhile, your eyes would be boiling/exploding. So why does diving work? Simple, the air inside you compresses until there is no longer a pressure differential. The same happens in a vacuum, but by the time there is no longer a pressure differential in this case you would be dead.
But when you look at the numbers another way, calculating an individual person's likelihood of being involved in a fatal automobile accident in his or her lifetime, the percentages come out very close to zero.
:-)
Well, it is zero. You weren't involved in a fatal accident until you die, in which case it wasn't during your lifetime.
This system could be taken out with a fairly ordinary power drill and a long bit, drilling one hole in the tube. Even just the partial loss of vacuum would be a huge problem for such high speed carriages.
Taking out a bridge with ordinary tools would be quite a bit harder (esp since its users could most likely see you doing it).
Using only metric simply garuntees that you'll be off by convenient powers of 10 when you violate your interfaces. If the interface checked the units, then in either case you would have easily found a problem.
/10, etc which could easily be forgotten and lead to bugs. We still have a platform that uses int usec for times, and occasionally still get bugs due to this (telling the difference between 10000000 and 1000000 while writing or scanning over code is easy to mess up).
In the robotics software my group writes, we've really had to drill it in to people to use the same metric unit (mm, sec, everywhere). In the past some of the modules differed, and had all sorts of ulgy *10
In short, merely using metric doesn't garuntee the nonexistence of unit bugs, and the decrease in bugs due to things like inch/cm is offset by the fact that people just *know* "we all use metric so it will work" and thus are lazy in checking the order of the units "{mm,cm,m,km} is obviously the most logical since it makes the most sense for my module!". Common interfaces still need to be established and conformed to, that is the root problem.
He does give noise measurements right here. Or else he's really, really, fast at responding to criticism.
There's no such thing as the "Sony Pino". Pino is from Kitano Symbiotic Systems Laboratory
The Sony robot is the SDR-4X. It's a lot more capable (it can actually walk right now), and a lot more expensive (though neither robot is being sold yet).
Too bad the user has to be running VirusScan or some other badly coded service for it to work. Funny how clueless nix users can be about their "enemy".
You seem to have missed the *whole point* of the original article. Go read it, and you'll find out any program with a text box will do. Poor coding has nothing to do with it, the bugs are in the Win32 API itself.
The strange thing is, he's not actually arguing that at all, and after criticizing things like the DMCA, goes on to say things such as:
Fortunately the technology itself - in the form of trusted computer architectures, secure networks and digital rights management - can be used to rescue the Net from US control.
I'm glad DRM can "rescue" us from the DMCA and evil corporate influence... They must take back the free network they had... and regulate it. yeah...
Well, it seems to be at least somewhat related to the Software Engineering Institute / CERT, etc. which is fairly separate from the rest of CMU, and a lot more large-software-corporation oriented than many of the research groups are. Of course, one of the nice things about a university is that it can be large enough to support both open communities and proprietary ones. Instead of trying to *fix* this one however, why not just support other open security projects, such as Ballista (which incidentally is also at CMU).
You mean these movies.
We have a new omni design as well, and it looks like the finalized robot (with all the junk on it) is going to be pretty fast. We'll have to watch out for the FU-Fighters, I hear they have an 8m/s (unconfirnmes) kicker they used at the German open. Of course kicking speed was not their problem in the past, but still... This year is going to be fun.
:P (we didn't have ours yet). Are the new robots strong enough you can you stand on their covers like the 2000 robots?
P.S. It was no fair not showing us your new robots when we visited
That would be the FU-Fighters 1999 team. It was a spinning 5mm thick aluminum plate, which is pretty large considering the robot itself was about 10cm x 12cm. They could kick the ball at up to 6m/s. They turned it down later in the competetion, but it was not banned (more to keep from getting penalties than other team's anger).
The design lives on still in a toned down version on the current FU-Fighters team and others have copied it too. It's affectionately known as a "spinning blade of death" kicker.
(Note: I'm sitting in a RoboCup lab right now, so IMNSHO:)
:) There are a lot of strong teams in the league though so we'll have to see...
:) It sure did that for me, even though I've been programming them for several years.
I don't think the favorite is going to be Iran this year, but more likely the Phillips professional team, which won the German Open this year. That said, I wish people would realize there are 4 leagues, not just the middle size league, with different robots and different favorites in each. In the Sony Legged league, UNSW has dominated, though we came in second
In the small size, I'd say the favorites are last year's winner LuckyStar II from Singapore, and Big Red from Cornell University. FU-Fighters is also a pretty strong team. Our team (CMU) hopes to do a lot better this year in the small size league. We won in '97 and '98, but haven't done too well since then.
I don't know to much about the simulation league so I won't bother to comment. Finally, a personal plug: See a video from the vision system of a Sony legged robot here. It'll give you more respect for how hard a problem this is
None of it has anything to do with intelligence. Again, to be intelligent, a program must not only be able to learn but it must learn anything and everything, not just a limited domain environment. But that is not all. It must be motivated and adapt to reward and punishment.
Any of course, a human *can* learn "anything and everything"? Don't make me laugh. Humans are actually quite limited, just in different ways than computers. There are tons of things that already come up in AI that a human can't learn.
This doesn't seem to stop people from saying a computer isn't intelligent until it can learn everything, or in other words be far superior to a human.
Netscape looked like it was going to die, so they started Mozilla, which I'm using happily right now. there aren't too many examples, but there are some...
Why did I buy an Nvidia video card? Well, because I actually did research
rather making vague comparisons. As always, a well set up system with the best drivers will beat one where the person didn't even get agp enabled. If not, then explain why I should trust your opinions over a well known review site...
Yes, their lack of source is annoying. But the fact that they care about Linux as a platform (more than any other card manufacturer) means a lot to me. I know there are people at Nvidia paid to write those drivers, and as long as they have the fastest hardware too, they're likely to always be ahead of any other hardware/driver combination in Linux.
Ethernet is not expensive. They could probably add it for $10 or less. As others have pointed out, they'd save the money that it costs them for dialup for users that already have broadband. They could even do cooler "enhanced functionality" things with the software since it'll have a continuous connection instead of an intermittent one.
That said, I guess I can understand why it doesn't have an ethernet port yet. I expect to see one reasonably soon though.
One of the things that makes the X-box cool is that it does have an ethernet port built in; It allowed my brother and me to play Halo with others over his cable modem for free - That's a definite edge over the PS/2, and that feature didn't cost them anything besides ethernet support (they didn't even write the tunnel software, it was 3rd party).
I agree completely. Working with robots and attending public competitions every year (I'm on a RoboCup team), the topic of beam robots and Tilden's work invariably comes up at some point. People need to know that while interesting, this work doesn't scale. Right now, nothing scales very well; But competing to see how efficiently you can mimic something dumber than a sea-monkey doesn't make much of a difference. Tilden probably understands this, but of course he never says this in interviews, which is a shame.
So, I usually tell people this: Rodney Brookes' robots did pretty much the same thing in the 80s. These analog robots are cheaper and simpler, but no smarter (and usually dumber). I will believe they are the future of robots when they can beat the best computer controlled robot team at any nontrivial robotics competition (AAAI, RoboCup, Mirosot, etc.)